As an search engine marketers, we analyze quite a lot of data looking for trends in the way people consume content online. Using the proper verbiage or titling a page one way or another can make or break a campaign. After years of doing this kind of work you really start to have an appreciation for just how much data Google has access to. It’s truly incredible. Looking backward at past trends you can often make very accurate predictions about the future. Maybe even predict with great accuracy who the next United States president will be?!
One tool that is particularly capable of putting mass amounts of data at your fingertips is Google Trends – trends.google.com. If you haven’t ever used this tool before, it essentially pulls search data and crunches it into a simple 1-100 number that represents relative interest in a certain topic or search entity over time. It’s great for quickly comparing search terms and gauging popularity.
The other day I was curious about how much interest there was in the current presidential candidates relative to each other. I found some pretty interesting information. I was excited to see that interest in Gary Johnson has been skyrocketing since September, however unfortunately, it’s only a drop in the bucket when compared to Hillary Clinton and Donald Trump. I also found that there has been consistently more interest in one candidate over the other. Although this tool can’t tell me if the interest is negative or positive.
This information made me curious though about what Google trends might be able to show me about past presidential races. What I found was truly astounding!
Google Trends data goes back to 2004. Because of that I was only able to pull data on the last 3 elections. In the case of those elections though, Google trends accurately predicted the winner months before the actual election! In the last 3 elections, simply by looking at search data, one could have accurately predicted who would win. This alone was incredible but I wanted to know how closely search data was to the recorded electoral vote.
I opened up Wikipedia and navigated to the Bush/Kerry election, the Obama/McCain election, and the Obama/Romney election. By dividing the percentage of electoral votes the losing candidate received by the percentage the winner received, I came up with the a 2 digit number indicating by how much one one candidate won over the other.
I then went back to Google Trends and divided the relative interest the losing candidate had in October, the month before voting, by the relative interest the winning candidate had in October in Google’s search engine. Because Google Trends shows the relative interest of one search entity compared to another, I was able to get another number that indicated by how much one candidate was popular in search engines compared to the other.
When I compared one set of numbers against the other I almost couldn’t believe what I saw. Search data was off from the recorded electoral vote by a maximum of only 10%.
- Electoral Vote: 0.86
- Search Interest: 0.76
- Electoral Vote: 0.47
- Search Interest: 0.45
- Electoral Vote: 0.62
- Search Interest: 0.71
Pretty incredible right? Google not only predicted the winner of each election, but in the case of the Obama/McCain race, Google predicted the winner by nearly the exact percentile!
Now, going back to the future here, in regards to the current election it is clear that Mr. Trump is going to be the next president of the United States. For better or for worse it seems to be inevitable. We can see that he has clearly more interest in Google as compare to his competitor. It doesn’t seem to matter whether the interest is positive or negative.
Let me know what you think in the comments below!